Operations Management 3QC3 February 1 , 2014
Chapter 4: Forecasting
Disney looks at two numbers – actual attendance and forecasted attendance. Disney’s 5 year
attendance forecast yields just a 5% error on average. Its annual forecasts have a 03% error.
What is Forecasting?
Making good estimates is the main purpose of forecasting. Good forecasts are an essential part of
efficient service and manufacturing operations.
Forecasting: The art and science of predicting future events.
There is not one forecasting techniques that works for everyone; they are rarely if ever perfect
and they are costly as well as timeconsuming.
Forecasting Time Horizons
A forecast is usually classified by the future time horizon that it covers.
1. Shortrange forecast – time span of up to one year, but usually less than 3 months. Used
for planning purchasing, job scheduling, workforce levels, job assignments, and
2. Mediumrange forecast – generally spans from three months to three years. Useful in
sales planning, production planning, budgeting, cash budgeting, and analysis of various
3. Longrange forecast – generally three years or more, they are used in planning for new
products, capital expenditures, facility location or expansion, and search and
Differences b/w short and medium/long range forecasts
• Intermediate and LR forecasts deal with more comprehensive issues
• ST forecasting usually employs different methodologies than LT forecasting. Techniques
such as moving averages, exponential smoothing, and trend extrapolation are used in ST.
• ST forecasts tend to be more accurate than LT.
After each sales period, forecasts should be reviewed and revised.
The Influence of Product Life Cycle
Most successful products pass through 4 stages: 1) introduction, 2) growth, 3) maturity, and 4)
Products in the first two stages need longer forecasts than those in maturity and decline.
Types of Forecasts
1. Economic forecasts – address the business cycle by predicting inflation rates, money
supplies, housing starts, and other planning indicators. MediumLR forecasts
2. Technological forecasts – concerned with rates of technological progress. LR forecasts
3. Demand forecasts – projections of a company’s sales for each time period in the
planning horizon / demand for the company’s products or services. Also called sales
forecasts, it drives a company’s production, capacity, scheduling system, etc.
The Strategic Importance of Forecasting
The forecast is the only estimate of demand until actual demand becomes known.
Product demand forecast impacts the following activities:
1. Human resources [Type text] [Type text] [Type text]
3. Supplychain management
7 Steps in the Forecasting System
1. Determine the use of the forecast
2. Select the items to be forecasted
3. Determine the time horizon of the forecast
4. Select the forecasting model(s)
5. Gather the data needed to make the forecast
6. Make the forecast
7. Validate and implement the results
Both product family and aggregated forecasts are more accurate than individual product
Forecasting Approaches ▯2 options
Quantitative forecasts – It uses a variety of mathematical models that rely on history data
and/or associative variables to forecast demand.
Qualitative forecasts – Incorporate such factors as the decision marker’s intuition, emotions,
personal experiences, and value system in reaching a forecast.
Overview of Qualitative Methods – 4 techniques
1. Jury of executive opinion: Uses the opinions of a group of highlevel experts or managers,
often in combination with statistical models, and combines the ideas to arrive at a group estimate
2. Delphi method: Includes 3 groups: decision makers (usually 510 experts, who will be
making the actual forecast), staff personnel, and respondents.
3. Sales force composite: Each salesperson estimates what sales will be in his or her region.
These are then reviewed and combined at the district and national levels to reach an overall
4. Consumer market survey: Solicits input from customers or potential customers regarding
future purchasing plans. However, this can suffer from overly optimistic forecasts that arise from
Overview of Quantitative Methods
5 quantitative forecasting methods, fall into 2 categories
Category 1: TimeSeries Models
• Naïve approach
• Moving averages
• Exponential smoothing
• Trend projection
Category 2: Associative Model
• Linear regression st
Operations Management 3QC3 February 1 , 2014
Time series – models predict on the assumption that the future is a function of the past, aka it
uses a series of past data points to make a forecast. These data points are evenly spaced.
Predictions are based only from past values.
Decomposition of a Time Series
Analyzing time series means breaking down past data into components and then projecting them
forward; it has 4 components:
1. Trend – gradual upward or downward movement of the data over time.
2. Seasonality – data pattern that repeats itself after a period of days, weeks, etc. 6 common
seasonality patterns pg. 110
3. Cycles – patterns in the data that occur every several years; usually tied into the business
cycle and are of major important in ST analysis and planning.
4. Random variations – “blips” in the data caused by chance and unusual situations
Naive approach: The most costeffective and efficient objective forecasting model. It assumes
that demand in the next period is equal to demand in the most recent period.
Moving averages: Uses a number of historical actual data values to generate a forecast. Useful if
we can assume that market demands will stay fairly steady. It uses an average of the n most
recent periods of data to forecast the next period. I.e. 4 month moving average: sum the demand
in path 4 months and divide by 4.
Moving average = sum of demand in previous n periods / n
When a detectable trend/pattern is present, weights can be used.
Weight moving average = sum of (weight for period n)(demand in period n) / sum weights
Both simple and weighted moving averages are effective in smoothing out sudden fluctuations in
the demand pattern to provide stable estimates.
Moving averages do present 3 problems:
1. Increasing the size of n does smooth out fluctuations better, but it makes the method less
sensitive to real changes in the data
2. Moving averages cannot pick up trends very well. They will always stay within past
levels and will not predict changes to either higher or lower levels, they lag the actual
3. Moving averages require extensive records of past data
Though both moving averages and weighted moving averages lag, weighted moving averages
react more quickly to demand changes.
Exponential smoothing: A weightedmoving average forecasting method in which data points
are weighted by an exponential function. It involves very little record keeping of past data.
New forecast = last period’s forecast + α (last period’s actual demand – last period’s forecast)
Where, α is a weight, or smoothing constant: the weighting factor used in an exponential
smoothing forecast, number between 0 and 1. It is chosen by the forecaster.
The smoothing constant is generally in the range from .05 to .5 for business applications. When it
reaches 1, then all older values drop out and the forecast becomes identical to the naïve model.
Selecting the smoothing constant: high values of the smoothing constant are chosen when the
underlying average is likely to change. Low values are used when the underlying average is
fairly stable. [Type text] [Type text] [Type text]
Measuring Forecast Error
Comapre forecaster and actual values. It tells us how well the model performed against itself
using past data. Forecast error:
Forecast error = Actual demand – Forecast value
= A t F t
F t the forecast in period t
A t the actual demand in period t
Several measures are used to calculate forecast error. 1) MAD 2) MSE 3) MAPE
Mean absolute deviation (MAD): A measure of the overall forecast error for a model, computed
by taking the sum of the absolute values of the individual forecast errors (deviations) and
dividing by the number of period of data (n)
MAD = sum of (actual – forecast) / n
The smaller MAD is preferable.
Mean squared error (MSE): average of the squared differences between the forecaster and
MSE = Sum of (forecast errors) / n
Again, you want to minimize MSE. It exaggerates errors because it squares them.
A drawback of using MSE is that it tends to accentuate large deviations due to the squared term.
Typically indicates that we prefer to have several smaller deviations rather than even one large
Mean absolute percent error (MAPE): The average of the absolute differences between the
forecast and actual values, expressed as a percentage of actual values.
MAPE = [100(actualforecast)/actual] / n
MAPE = [100(error/actual)] / n
A problem with MAD and MSE is that their values depend on the magnitude of the item being
forecast. If the forecast item is measured in thousands, the MAD and MSE values can be very
large, to avoid use MAPE.
Exponential Smoothing With Trend Adjustment
Simple exponential smoothing fails to respond to trends. New formula:
Forecast including trend (FIT) = expotentially smoothed forecast (F) + exponentially tmoothed
trend (T) t
This procedure requires two smoothing constants: α for the average, β for the trend.
F t α (Actual demand last period) + (1 α)(forecast last